The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides tools for searching the web and fetching the full text of webpages, which involves processing untrusted external data that could contain malicious instructions.
Ingestion points: The blockrun_exa tool outputs from the search, answer, and contents endpoints contain raw data from the web.
Boundary markers: The instructions do not specify any delimiters or safety warnings for the agent to use when interpreting the fetched web content.
Capability inventory: The skill uses the BlockRun SDK to perform network requests and manage wallet-based payments for search operations.
Sanitization: There is no mention of sanitizing or filtering the external content before it is provided to the LLM context.
[EXTERNAL_DOWNLOADS]: The skill instructions specify the installation of an external Python package.
Package: blockrun-llm (installed via pip install blockrun-llm).
Context: This is a vendor-provided SDK required for interacting with the search API and managing payments.

exa-research